Seyfert Galaxies Encyclopedia of Astronomy & Astrophysics

eaa.iop.org DOI: 10.1888/0333750888/1584

Seyfert Galaxies Mark Whittle

From Encyclopedia of Astronomy & Astrophysics P. Murdin

ISBN: 0333750888

Institute of Physics Publishing Bristol and Philadelphia

Downloaded on Thu Mar 02 23:38:01 GMT 2006 [131.215.103.76]

Terms and Conditions Seyfert Galaxies E NCYCLOPEDIA OF A STRONOMY AND A STROPHYSICS

Seyfert Galaxies are thought to be instrumental in funneling gas down to the nuclear regions where it can then fuel the central black In many respects Seyfert galaxies are similar to any hole. other galaxy, except their nuclear regions show unusual properties: they can be much brighter and their spectra Discovery, classification and detection show strong emission lines of high excitation. Such nuclei The early identification of nearby Seyfert galaxies rested are called ‘active’, and so Seyfert galaxies belong to the on two observational properties—their unusual nuclear wider category of ‘ACTIVE GALAXIES’ or, more specifically, spectra and their unusual nuclear brightness. While they have ‘ACTIVE GALACTIC NUCLEI’ (AGN). Seyferts were, studying galaxy spectra for his PhD work in 1908, Fath in fact, the first type of active galaxy to be discovered. noted strong emission lines in the spectrum of the nucleus In the zoo of AGNs the power of the nuclear activity of NGC 1068. These observations were confirmed in 1917 spans a wide range. Within this zoo, Seyferts show by Slipher and by 1926 Hubble had added two more intermediate levels of activity, being more powerful than galaxies with similar spectra: NGC 4051 and NGC 4151. In LINER galaxies but less powerful than QUASARS. Seyferts 1932 Humason noted that NGC 1275 had a bright starlike also join the majority of active galaxies in being ‘radio- center, but it wasn’t until 1943 that CARL SEYFERT recognized quiet’, with radio luminosities 103–104 times weaker than a distinct class of galaxies with unusually bright and the ‘radio loud’ category, which includes RADIO GALAXIES concentrated nuclei, and studied in detail six of the 12 and radio quasars. then known cases (NGC 1068, 1275, 3516, 4051, 4151, Over the past 20 years much effort has gone into 7469), leading to the group’s name: Seyfert galaxies (see finding Seyfert galaxies amongst the more normal galaxy figure 1). Seyfert’s spectra showed emission lines which population. Efficient ways to detect Seyferts include were not only unusually strong, but also unusually wide. looking specifically for galaxies with bright UV emission, Interpreting the width of the spectral lines as Doppler bright x-ray emission, or unusual far-infrared colors, since shifts caused by motion of the ionized (line emitting) gas these are common characteristics of Seyferts. The fraction implied gas velocities in some objects of several thousand − of galaxies with active nuclei depends on the level of km s 1, far higher than velocities found in normal galaxies. activity: moderate luminosity Seyferts make up ∼ 1–2% of A further subtlety Seyfert noticed was that in some objects nearby bright galaxies, lower luminosity Seyferts make up the hydrogen lines were broader than the other lines. ∼ 10%, while LINERs make up ∼ 40%. Currently, about Little work was done until the 1960s, when it became 800 Seyferts are known, though the number continues to clear that Seyfert galaxies shared a number of properties rise. with the then recently discovered quasars. At this time, From a wide range of studies, a picture has emerged of quasars posed a serious theoretical puzzle. While their the structure of the central regions of Seyferts. Briefly, the high REDSHIFTS implied great distance and therefore high gravitational field of a supermassive (106–109 M) black luminosity, their brightness had been found to vary from hole provides the ultimate energy source, as gas falls month to month, indicating that the region producing inwards via a luminous accretion disk (see SUPERMASSIVE all this energy was only a few light months across—tiny BLACK HOLES IN AGN). A small region of fast moving clouds compared to the rest of the galaxy. The puzzle of how so surrounds this central engine, spanning a few light days much energy could emerge from so small a volume was or weeks, while a larger region of slower moving clouds sufficiently acute that some astronomers even questioned extends out a few hundred to a few thousand light years. whether the quasar redshifts were true indicators of In some cases, oppositely directed collimated jets of low distance, leading to the so-called ‘redshift controversy’. density gas plow into the surrounding galaxy, accelerating Although this controversy has since been resolved in the interstellar medium and creating jet-like or linear favor of large distances it might never have been as radio structures. In at least some Seyferts, dense gas and problematic if Seyfert galaxies had been better studied dust shroud the innermost regions, hiding them from our in the early 1960s, since they are clearly low redshift direct view, though the nuclear radiation escapes in other counterparts to the quasars. Not only are their spectra directions to light up gas and dust residing much further basically similar, but in 1968 NGC 4151 was found to vary in out in the galaxy. brightness, confirming a highly compact nuclear energy The conditions that give rise to Seyfert activity source. As more Seyferts and quasars were discovered, are still under debate. While Seyferts are found in a their separation in luminosity narrowed and finally even wide range of galaxy types, they are more commonly overlapped. Continuity had been established and, for found in reasonably luminous early-type spirals (e.g. most astronomers, the reality of the quasar distances and types S0/a, Sa, Sb in the Hubble classification scheme). luminosities was no longer in doubt. It seems, therefore, that prerequisites favoring Seyfert In 1974, Khachikian and Weedman identified two activity include a massive galaxy bulge and the presence types of Seyfert galaxy on the basis of the widths of the of an interstellar medium. There is also evidence that nuclear emission lines. Figure 2 shows example spectra Seyferts prefer galaxies with an asymmetric gravitational of each Seyfert type. While spectra of type 2 Seyferts field arising from a bar or other distortion, possibly have a single set of relatively narrow emission lines, in induced by a nearby companion galaxy. Such distortions the spectra of type 1 Seyferts the hydrogen and helium

Figure 1. Three photographs of NGC 4151 taken with increasing exposures (a, b, c) showing its unusually bright nucleus—an important characteristic used in the early identification of Seyfert galaxies. (Taken from Morgan W W 1968 Astrophys. J. 153 27) lines have an additional much broader component. In Seyferts was that of Markarian and collaborators at the the simplest cases, the broad component is either absent Byurakan Observatory inArmenia, during the period 1962 (Seyfert 2) or strong and dominant (Seyfert 1). With to 1981. This survey used a technique known as Objective better data it became clear that there is a wide range in Prism Spectroscopy to find galaxies with unusually blue the relative strength of the broad and narrow emission continuum emission. The telescope was a 1.3 meter lines, and this led Osterbrock in 1981 to refine the Seyfert Schmidt telescope with a thin prism bolted over the classification by introducing intermediate types. A good entrance aperture. This combination gives relatively wide place to establish the classification is with the hydrogen Hβ field photographs in which each star and galaxy appears as line at rest wavelength 4861 Å. As the broad component a small spectrum. In this way, the spectra of many galaxies of Hβ becomes weaker relative to the narrow component, can be studied quickly, and those which appear unusual the Seyfert type changes from 1 to 1.2 to 1.5 to 1.8. For noted and listed. The survey covered ∼ 10 000 square Seyfert 1.8 galaxies, weak broad wings are just visible at degrees and yielded 1500 galaxies with blue continua, of the base of Hβ while for Seyfert 1.9 galaxies they are only which ∼ 10% were Seyferts (the remainder were mainly visible on the Hα emission line at 6563 Å. In practice, STARBURST GALAXIES, blue because of the presence of young these Seyfert sub-types have not been formally defined but massive O and B type stars). Because of the emphasis instead give an overall indication of the degree to which on blue continuum, the Markarian Seyferts contain a the broad component is present. relatively high proportion of Seyfert 1s (about 50%) and Although there are a number of Seyfert properties that rather luminous Seyfert 2s. depend on Seyfert type, perhaps the most noticeable is Since x-ray emission is common in active galaxies, the ‘white light’ continuum coming from the nucleus. In x-ray surveys provide an effective means to find Seyferts. normal galaxies the nuclear continuum is mostly starlight, Starting in the late 1970s surveys have included those of the but in active galaxies there can be additional emission, Ariel V, HEAO-1, Einstein and ROSAT satellites. Some of often blue in color and with few or no absorption features. these were all-sky surveys, some included deeper pointed This component is much stronger in Seyfert 1s, weaker observations, some were sensitive to hard x-rays (2–10 in Seyfert 2s, and may even be undetectable in the lowest keV) and others to soft x-rays (∼ 0.2–4 keV). In general, luminosity Seyfert 2s. Although we refer here to the optical x-ray surveys tend to find Seyfert 1s since their nuclear continuum emission, this component extends into the UV continuum emission is stronger than in Seyfert 2s. and x-ray, and so emission in these spectral bands can also In 1982, the IRAS satellite provided a huge list of be quite different for Seyfert 1s and Seyfert 2s. galaxies which were relatively bright in the far infrared Bearing in mind the principal features which (25–100 µm). Although bright infrared flux by itself is distinguish Seyfert spectra from normal galaxy spectra— no guarantee of Seyfert activity, the ‘color’ of the infrared strong broad emission lines and a blue continuum—we emission can help distinguish between Seyfert and normal can now follow the history of how Seyferts have been galaxies: normal galaxies have colder dust (∼ 30 K) identified over the past 40 years. than Seyfert galaxies (∼ 100–300 K) in which nuclear As outlined above, the first Seyferts were discovered activity provides an additional heating source. Thus, essentially by accident when a galaxy spectrum was noted IRAS-selected galaxies with ‘warm spectra’ have yielded as being unusual. The first survey which identified many many Seyferts.

A catalog of all known Seyferts has been maintained by Veron-Cetty and Veron, and at its last edition (1998) about 790 Seyferts were listed with redshifts < 0.1 and brighter than 17m in V.

Overall structure and geometry In this section we give a brief outline, without justification, of the basic structures found in Seyfert galaxies; brief because similar material is given elsewhere in this encyclopedia and also because more detailed descriptions of some of the structures are given in subsequent sections. Figure 3 gives a cartoon of the structures nested from innermost to outermost. The region sizes probably scale with luminosity, larger for quasars and smaller for low luminosity Seyferts and LINERs. The values given are rough guides for the case of moderate luminosity Seyferts. The ultimate origin for the energy which drives nuclear activity is thought to lie in the gravitational field of a central supermassive black hole of mass ∼ 107–108 M. Gas falls into the black hole via an accretion disk, which radiates powerfully across much of the electromagnetic spectrum. The disk may span from a few Schwarzschild radii to a few thousand Schwarzschild radii, probably with x-ray emission coming from the inner parts, UV and optical from further out. It is this radiation which floods out into the galaxy and ionizes any surrounding gas, causing it to radiate emission lines. The innermost region of ionized gas resides a few light days from the center, and contains gas with density ∼ 109–1012 atom cm−3 moving with speeds of a few thousand km s−1. Doppler shifts therefore lead to broad emission lines— hence the name: Broad Line Region (BLR). Outside the BLR, on scales from one to a few pc lies thick dense gas, possibly in the form of a torus. This gas blocks radiation coming from the innermost regions, which can Figure 2. Spectra of three Seyfert galaxies illustrating the only escape out of the equatorial plane along the torus difference between Seyfert 1, 1.5 and 2 classes. Principal axis. This partially collimated radiation field emerges emission lines are labeled, with square brackets indicating further into the galaxy to ionize gas residing in a region forbidden lines. (Adapted from Osterbrock D E 1984 Quart. J. R. out to a few kpc. In this region, the gas has a lower Astron. Soc. 25 1.) − density ∼ 102–106 atom cm 3 and moves with speeds of a few hundred km s−1, leading to narrower emission Since Seyferts belong to the radio-quiet group of lines—hence the name: Narrow Line Region (NLR). active galaxies, radio surveys at high flux levels tend not Because of the partially collimated radiation field, the to find Seyferts, but rather radio galaxies. Deeper radio NLR often appears elongated or even biconical, with surveys, however, do pick up more Seyferts but along axis matching that of the central radiation field. When with many star-forming galaxies. On further study, the a radio source is present it is often elongated, aligned presence of a compact core with flatter spectral index with the NLR axis, and has approximately the same size. increases the chance that the galaxy is a Seyfert. The radio source emerges from the inner regions as a Perhaps the most reliable method of finding Seyferts, bipolar flow, accelerating and possibly shocking some of though also the least efficient, is optical spectroscopy of the gas in the NLR. The entire NLR and associated radio galactic nuclei. In the early 1990s the CfAredshift survey of structure usually fill a region a few kpc across within about 2000 galaxies brighter than about magnitude 14–15 the bulge of the host galaxy, but randomly aligned with found about 2.5% to be Seyferts with an even mix of Seyfert respect to obvious galactic structures and orientation. If 1s and 2s. Using smaller apertures and higher quality data, there is a more extended gaseous component, it can be the deep Palomar survey of the nuclei of nearly 500 bright illuminated by the central ionizing radiation field, leading galaxies from the RSAcatalogue yields about 10% Seyferts, to faint extended emission, sometimes showing conical or including many of lower luminosity. In the near future, the biconical form. This is the so-called Extended Narrow Line SLOAN DIGITAL SKY SURVEY will surely find many more. Region (ENLR).

BH & Accretion Disk BLR Dense ISM Torus / Lanes / Random ?

RAD ~ 3 –1000 RS RBLR ~ few lt. days R ~ 1 – few pc

Nested structures

within a kpc R ~ 1 – few

Radio Seyfert galaxy Source

Host galaxy & ENLR NLR

Figure 3. Cartoon illustrating the nested structures thought to be present within Seyfert galaxies.

Inherent in this picture is a lack of spherical symmetry, elongated, with a subset showing sharp edges and even rooted in the anisotropic emergence of both radiation conical or biconical form. These patterns are consistent and radio emitting flows. The origin of the anisotropy with dense nuclear gas casting shadows in a strong might arise in either of two ways: (a) an intrinsic nuclear central ionizing radiation field. When visible, the cone ◦ ◦ axis perhaps defined by the black hole spin and/or the opening angles are ∼ 40 –80 , suggesting ∼ 25% of the accretion disk, and (b) partial obscuration by dense gas surrounding sky sees the nuclear regions. Good examples which blocks radiation in some directions but not in of this are found in NCG 1068, NGC 5252, MKN 78, MKN others. For Seyferts, partial obscuration is thought to 573. be important. The simplest geometry envisioned is a Extending this topic further, one can use spectroscopy thick disk or torus a few parsecs in size and aligned with of the ionized gas to show that it sees a significantly the central engine axis. More recently, images from the brighter central radiation source than we see. There are Hubble Space Telescope (HST) have shown that dust lanes two approaches. In the first, emission line strengths and filaments are common in the central few hundred from extended gas are used to infer both the gas density parsecs, particularly in Seyfert 2s, raising the possibility and its degree of ionization. Knowing the distance that less coherent obscuration on these larger scales might between this gas and the central source then allows one also be important. Either way, the fact that the nuclear to estimate the (ionizing) luminosity of the central source. radiation can only emerge along particular lines of sight This luminosity is often significantly greater than the raises an interesting possibility which was first noted in luminosity we see directly, along our line of sight. The 1978 by Osterbrock: namely, that Seyfert 2s might actually β be Seyfert 1s but oriented in such a way that our view to second approach uses the total H luminosity coming the central region is blocked. Stated slightly differently,the from all regions to estimate the total ionizing luminosity of same Seyfert galaxy might appear as a Seyfert 1 from some the central source. This luminosity is again often found to directions and a Seyfert 2 from others. This constitutes a be significantly greater than the luminosity we infer from ‘Unification’ scheme, since apparently different classes of our own line of sight to the nucleus. objects are ‘unified’ into a single class. The important topic The third and perhaps most dramatic piece of of unification arises in a number of contexts in the study of evidence involves taking spectra of light that was initially active galaxies and is discussed at length in ACTIVE GALAXIES: moving out of the nucleus in one direction but then got UNIFIED MODEL. Here we give a brief outline of the evidence reflected into our line of sight. This provides a sort supporting the particular case that many (though perhaps of periscope, allowing one to look around a corner and not all) Seyfert 2s are in fact hidden Seyfert 1s. into the nucleus from a different direction. The mirrors First, the emission line regions of Seyfert 2s, on both of the periscope are dust particles or electrons in the large scales (ENLR) and small scales (NLR), are often surrounding gas which scatter light in all directions, some

Copyright © Nature Publishing Group 2001 Brunel Road, Houndmills, Basingstoke, Hampshire, RG21 6XS, UK Registered No. 785998 and Institute of Physics Publishing 2001 Dirac House, Temple Back, Bristol, BS1 6BE, UK 4 Seyfert Galaxies E NCYCLOPEDIA OF A STRONOMY AND A STROPHYSICS of which comes towards us. Since the scattered light is of which ‘Seyferts’ are only one class. This situation has so faint compared to the normal light we need a way arisen for a couple of reasons: (a) nuclear activity seems to isolate it. Fortunately, the scattering process also to have some genuine range of character, and (b) historical induces polarization, and so a spectrum of the polarized development is never clean and different approaches have light shows just the scattered component. Remarkably, naturally focused on different manifestations of activity. for many of the brighter Seyfert 2s these polarized light In hindsight, it is now possible to group the various spectra show Seyfert 1 form: strong non-stellar continuum categories in a more or less natural way which depends on and broad hydrogen and helium lines. The classic two or three fundamental aspects of the activity itself: total discovery was for NGC 1068 by Antonucci and Miller in 1983, luminosity, radio luminosity and variability. While these but since then many other examples have been found. aspects are discussed in ACTIVE GALAXIES: OVERVIEW and ACTIVE A fourth approach is to compare the nuclear x- GALACTIC NUCLEI: VARIABILITY, here we give a brief overview ray emission of Seyfert 1s and 2s. Consistent with of the different classes of AGN, intending to locate Seyferts unification, soft x-ray spectra show significant absorption in their proper place. towards Seyfert 2s but little or none towards Seyfert 1s. Consider first the enormous range in total luminosity ∼ 39 −1 M ∼− Furthermore, at higher x-ray energies where absorption of AGN: 10 erg s ( B 10) at the lowest detectable ∼ 47 −1 M ∼− isn’t important, the spectra of Seyfert 1s and 2s are often end, up to 10 erg s ( B 30) at the highest. quite similar, with ratios of x-ray to optical flux much Seyferts span the low to intermediate range, overlapping higher for the Seyfert 2s, as expected if the optical emission somewhat on the high end with quasars. Since it is blocked but the hard x-rays are not. now seems that Seyferts and quasars form a continuous Finally, there is some evidence for the nuclear physical sequence, any attempt at a formal separation is obscuring material itself. When water masers are detected, somewhat arbitrary. Although a morphological criterion they tend to be in Seyfert 2s. These sources form in dense is usually adequate (quasars have point-like appearance) warm molecular gas with long paths in our line of sight, close scrutiny of low redshift quasars often reveals galactic consistent with an edge-on disk geometry. Also, infrared ‘fuzz’ in ground-based images or clear host galaxies in spectra are often dominated by thermal emission from HST images. For this reason, a luminosity criterion is warm dust. Although some of this originates on NLR sometimes chosen to delineate Seyferts and quasars. There M ∼− scales, some comes from much smaller scales. The fact that is no consensus, but threshold values of B 23 or L ∼ 44 −1 the total luminosity of many Seyferts, particularly Seyfert B 10 erg s are typical. While Seyferts and quasars 2s, is dominated by the thermal infrared is consistent with may be physically similar, differences do seem to occur the obscuration picture: much of the nuclear luminosity is with increasing luminosity, including longer variability absorbed by gas and dust whose principal mode of cooling timescales indicating larger emission regions, the apparent is in the far infrared. loss of the ‘type 2’ class, and some line strength changes Assuming, for the moment, that all Seyfert 2s contain including the relative weakening of both broad and hidden Seyfert 1s, then we can ascertain the extent of the especially narrow lines relative to the continuum. obscuration by comparing the relative number of Seyfert Turning to lower luminosities, although one of the 2s (obscured) with Seyfert 1s (unobscured). In practice, it lowest luminosity AGNs is in fact a Seyfert (NGC 4395, M ∼− is important (and difficult) to select an unbiased sample of B nuc 10) most of the low luminosity AGNs belong to Seyferts using a property that is not thought to depend on the ‘LINER’ class. This group was originally identified by orientation. Possibilities include radio flux, far-infrared Heckman in 1980 from a spectroscopic survey of a large flux, host galaxy magnitude, and narrow emission line number of bright nearby galaxies. A high fraction were flux, though none are ideal. With some uncertainty, the found to have weak emission with low overall degree ratio of Seyfert 2s to Seyfert 1s found in such samples is of ionization, hence the name: Low Ionization Nuclear roughly 2–5 to 1, suggesting ∼ 20–30% of the sky sees Emission Line Regions, or LINERs. The exact nature of an unobscured view of the nucleus, consistent with other LINERs is still unclear, with evidence for an active nucleus estimates based on the opening angle of emission line in at least ∼ 20%: a compact nuclear UV source seen in HST cones. images; a compact though weak nuclear radio source; or Whether all Seyfert 2s contain hidden Seyfert 1s is still weak broad line emission visible at the base of Hα.A an open question. Arguments against full unification of major uncertainty is the ionization mechanism for the gas. the two classes focus either on the absence of expected Possibilities include shocks or photoionization by either a properties (for example, most Seyfert 2s have lower weak nuclear UV source, very hot young stars or very hot polarization than expected), or the fact that properties evolved stars. which should be the same in Seyfert 1s and 2s are not Shortly after the identification of LINERs, a method (for example, host galaxy morphology, and possibly radio for classifying emission line regions was introduced by luminosity). Baldwin, Phillips and Terlevich and subsequently refined by Veilleux and Osterbrock. They used a number of Relation of Seyferts to other AGN diagrams which plotted the ratio of one pair of emission To the newcomer, it may seem that there are an annoyingly line fluxes against another pair. Figure 4 shows one of large number of names for different classes of active galaxy, these diagrams and nicely illustrates how three major

Peaked Spectrum (GPS) sources). It is quite likely that similar physical phenomena are occurring in the Seyferts and these compact powerful sources—both are embedded in a dense gaseous environment and probably interacting strongly with it. Thus, studying one may help understand the other. The fundamental reason why ∼ 5–10% of AGN become radio-loud is still not understood, though it is probably related in some way to the interesting fact that radio-loud and radio-quiet AGN tend to inhabit different types of host galaxy: ELLIPTICAL and SPIRAL GALAXIES respectively. It has been suggested that the radio-loud AGN have rapidly spinning (Kerr) BLACK HOLES which drive powerful jets along the spin axis, and these jets generate the powerful radio source. Conversely, slowly rotating (Schwarzschild) black holes cannot drive such jets and so remain radio-weak. The link between black hole spin and host galaxy type could arise from the fact that the elliptical hosts have formed from the merger of two spirals along Figure 4. One of several useful diagnostic diagrams which help with the merger of their two black holes to form a single distinguish between different classes of emission line object. rapidly rotating black hole. (Adapted from Peterson B M 1997 An Introduction to Active Finally, Seyferts have little relation to the highly Galactic Nuclei (Cambridge University Press), figure 2.3, p 26). variable ‘BLAZAR’ classes of AGN: BL Lac objects, OVV (optically violently variable) quasars and HPQs (high polarization quasars). These are all radio-loud objects and, emission region types occupy different places on the under the unified scheme for this category, are thought diagram; in this case Seyferts separate from LINERs to be seen looking directly down a powerful jet where around [OIII] λ5007/Hβ ∼ 3 while Seyferts and LINERs λ α ∼ relativistic Doppler boosting enhances the luminosity, separate from H II regions around [NII] 6584/H variability and polarization of the continuum. 0.6. These line ratio diagrams also provide a suitable place to compare observations with theoretical models. Black holes and continuum emission For example, the H II REGION sequence matches models It is worth noting that the evidence for black holes in of gas clouds with a range of metallicities photoionized Seyferts and other AGN is mostly indirect: large variable by radiation from OB associations with a range of ages. luminosity; relativistic jets with a stable axis; high energy Similarly, Seyferts and LINERs form a sequence that spectra. Ideally, one would study gravitational dynamics matches gas which is photoionized by a power law of near-nuclear stars or gas. While this has been possible spectrum with a range of intensities; the highest excitation in a number of normal or weakly active galaxies, it is very Seyferts correspond to the most intense nuclear radiation difficult for most Seyferts simply because of the bright field. The apparent continuity of Seyferts and LINERs on nuclear light and uncertain nature of the gas motions. With these diagrams supports the possibility that at least some many ongoing HST observations, the case is now building LINERs are simply low intensity Seyferts. for black holes at the centers of most (perhaps all) luminous Turning now to radio properties, Seyferts belong galaxies, with black hole mass a few per cent of the bulge to the majority (∼ 90%) of AGN in being radio quiet— mass. It would be ironic if the discovery of black holes in more luminous than normal galaxies but ∼ 103–104 most normal galaxies gave the most convincing evidence, times fainter than radio-loud AGN. The closest radio-loud by continuity, for their presence in AGN. counterparts to Seyferts have similar optical luminosity Two methods might provide a better approach to and are called Broad Line Radio Galaxies (BLRG) and detecting black holes in Seyferts. The first method is to Narrow Line Radio Galaxies (NLRG), with optical spectra measure the dynamics of water masers in the dense warm very similar (though not identical) to Seyfert 1s and gas close to the center. The classic case is NGC 4258 (not, Seyfert 2s respectively. At higher optical luminosity, we in fact, a Seyfert galaxy), which provides the best case have radio-loud quasars, and at lower optical luminosity to date for a supermassive black hole, but other possible we have radio galaxies whose spectra are often LINER- examples are increasing in number. The second method like. Although most radio galaxies have huge radio is to study the Fe Kα x-ray emission line profile at 6.4 sources which extend well outside the host galaxy, some keV. In several cases it is very broad and double peaked, have much smaller radio sources comparable in size to, suggesting its origin in a nuclear accretion disk at a few though much more powerful than, the Seyfert sources tens of Schwarzschild radii. An important goal will be to (these are called, depending on details, Compact Steep use profile fits to infer not only the mass of the hole but Spectrum (CSS), Compact Double (CD) or Gigahertz also its spin and orientation.

The energy released from matter falling towards debate, but it seems there are at least several contributions. the central black hole emerges in both bulk flows and Polarization observations indicate that part of the emission photons; the latter yields, either directly or indirectly, the is scattered from the hidden nucleus. The remainder observed nuclear continuum emission. Perhaps the most (sometimes called ‘FC2’, for Featureless Continuum 2) noteworthy aspect of AGN continua is their extremely may be a combination of near-nuclear star formation, wide spectral range, with approximately equal energy per Balmer continuum emission from the ionized gas, and decade of frequency from far infrared to 100 keV x-rays continuum emission from hot post-shock gas associated (five decades) with extensions to radio and γ -ray energies with the radio source. in about 10% of AGN (not usually Seyferts). To a first approximation, the continuum shape can be described Broad line region −α as a power law, fν ∝ ν , with index α ∼ 0–1, with Adefining characteristic of Seyfert 1 spectra is the presence several features superposed. These features include: (a) of broad emission lines, including the Balmer lines, He II a ‘Big Blue Bump’ (BBB) comprising a rise below ∼ 4000 Å λ4686 and He I λ5876 in the optical, and Lyαλ1212, C IV peaking in the UV or (unobservable) EUV and dropping λ1549, C III] λ1909 and Mg II λ2800 in the near UV. In the to rejoin the underlying continuum in the soft (∼ 0.5 keV) early 1980s it was found that the Balmer line strengths x-ray region (the ‘soft excess’); (b) an infrared peak rising in Seyferts and quasars form a single tight correlation from ∼ 1 µm, peaking at ∼ 25–60 µm and falling again at with the non-stellar continuum strength, confirming a ∼ 0.3 mm; (c) a rise at the highest energies ∼ 5–200 keV basic connection between the two: continuum emission with cut-off at even higher energies. photoionizes the BLR gas which then produces line The interpretation of these various components is radiation. An estimate of the gas density in the BLR still somewhat uncertain, though there seems to be a comes from the absence of any broad components on the general consensus. The underlying power law might forbidden lines. These so-called ‘forbidden’ emission lines be generated by the synchrotron self-Compton process: come from electron transitions which cannot occur via a population of relativistic electrons not only generates the fast electric dipole mode (which requires a quantum low energy photons via the synchrotron process but number change l =±1). With only slower modes of then inverse-Compton scatters these same photons up to transition available (e.g. magnetic dipole), the ions are higher energies. A related model for the x-ray power vulnerable to collisional deexcitation in a high density law involves inverse-Compton scattering of optical–UV gas where the mean time between collisions is small. n photons produced in the ACCRETION DISK by electrons in The ‘critical density’, crit, above which collisional de- a hot corona above the disk. The BBB is thought to excitation becomes important varies widely for different arise from thermal emission from the accretion disk, with lines, so one can bracket the gas density in the BLR by progressively higher energy photons coming from higher noting which lines have broad components and which do λ n ∼ temperatures at smaller radii. Although the overall BBB not. For example, while the [OIII] 4363 line with crit − feature can be nicely fitted by disk models, some expected 3 × 107 cm 3 has no broad component the semiforbidden λ n ∼ × 10 −3 disk signatures are not seen: a strong Lyman edge at 912 Å; CIII] 1909 line with crit 3 10 cm does, implying . high polarization; and significant phase delays between a gas density in the BLR somewhere in the range 107 5– . − flux variations at different wavelengths. For this reason 1010 5 cm 3 (see below for a refinement of this estimate). alternatives have been suggested, such as an optically thin In contrast to the forbidden lines, the so-called ‘permitted’ − bremsstrahlung component. The infrared peak, which can lines arise from fast (∼ 10 8 s) electric dipole transitions − be very strong in some objects (e.g. NGC 1068), is thought with correspondingly high critical densities ∼ 1017 cm 3. to arise in most cases from thermal emission from dust. Such lines, which include those due to recombination, are The rise longward of ∼ 1 µm suggests hot dust close to therefore easily produced in the BLR gas. its sublimation temperature at ∼ 2000 K, possibly located In the last decade, Seyfert variability studies have within ∼ 1 pc. At longer wavelengths (∼ 25–60 µm) the significantly helped develop our picture of the BLR. dust is cooler and probably located in or near the NLR. In these studies, the emission lines and continuum are Finally, the x-ray rise might come from reflection of the repeatedly measured over periods of many months. underlying power law off ‘cool’ gas, either the accretion Because of the finite size of the region, it takes time for disk or the dense obscuring torus. A byproduct of such a change in the central continuum brightness to reach the reflection is fluorescence of the Fe Kα emission line at ∼ 6.4 BLR gas and thereby cause a change in the emission line. keV, which can be quite strong. Allowing for light travel delays between the continuum The observed blue–UV continuum in Seyfert 2s source and BLR gas and between the BLR gas and us, deserves further comment since, as we have discussed, it is in principle possible to reconstruct the distribution the nuclear contribution should be hidden behind an of line emission and velocity field throughout the BLR. obscuring torus. Why, then, do many Seyfert 2s have a In practice, this full ‘reverberation mapping’ has not yet sufficiently strong UV excess to show up in the Markarian been possible, but instead simple time delays have been survey? The reason is that much of the blue–UV light measured in the response of each emission line to the in Seyfert 2s is extended, coming from the inner few continuum, and these delays indicate an approximate kpc. The origin of this continuum is still under some region size. The most well studied object is the Seyfert

h L ( , )1/2 h 10 0 λ 1350 Å 40 light days, where 0 is the Hubble − − constant in units of 100 km s 1 Mpc 1 and Lλ(1350 Å, 40) is the continuum luminosity at 1350 Å in units of 1040 erg s−1 Å−1. Since, roughly speaking, quasars are ∼ 100 times more luminous than Seyferts, they have BLRs which are ∼ 10 times larger, perhaps a few light months to a light year across. In the 1980s photoionization models were developed which aimed to reproduce the observed emission line strengths. In general, these calculations consider a power law (or AGN-like) continuum impinging on a thick slab of gas. At a given depth in the cloud, ionization and thermal equilibrium conditions are evaluated, together with the emitted line spectrum. Absorption of the incoming ionizing continuum is calculated and the modified continuum enters the next zone deeper in the cloud. The calculation repeats for ever deeper zones in the cloud until the gas is sufficiently cool and/or neutral that no more emission is produced. Overall, the ionization degree drops into the cloud, becoming suddenly more neutral at a depth corresponding to the classical Strömgren depth (where the total number of ionizing photons is balanced by the total number of hydrogen recombinations). However, unlike H II regions ionized by hot stars, AGN continua also have a high energy component which penetrates beyond the Strömgren depth to create an extended partially ionized zone in which many low ionization lines are produced. Usually, the emergent spectrum from a single cloud is compared to observations, Figure 5. Variability in NGC 5548 over a period of 280 days in though more recently cloud ensembles spanning a range 1988–89. Left panels show changing fluxes in the continuum of conditions have been considered in an attempt to match (upper two) and emission lines (lower three), while the right panels show the cross correlation of each flux curve with the UV the stratified nature of the BLR. continuum (uppermost left panel). Note that while the optical Several parameters help define these photoionization and UV continua vary together, the Lyα,CIVλ1550 and Hβ calculations. The most important is the ‘radiation emission lines lag behind by increasing amounts, Hβ being parameter’, defined as the ratio at the cloud front of the delayed by about 20 days. These lags indicate the size of the BLR ionizing photon density to the hydrogen density: U = (about 20 light days) and the fact that it is radially stratified, with N /n N = Q/ πr2c Q ion H, where ion 4 where is the number C IV produced preferentially at smaller radii and Hβ at larger of ionizing photons produced by the central source per radii. (Taken from Peterson B M 1997 An Introduction to Active r Galactic Nuclei (Cambridge University Press), figure 5.6, p 86). second, and is the distance to the cloud. Other relevant parameters are gas density, cloud thickness, ionizing spectral shape and element abundances. Optimum fits U ∼ . n ∼ 11 1.5 galaxy NGC 5548, for which delays of ∼ 2–10 days suggest that for Seyfert 1s, 0 04 and H 10 − are found for the highest ionization lines (e.g. He II λ1640, cm 3. Estimating cloud thickness is more difficult. The CIVλ1549) and ∼ 10–20 days for the lower ionization fact that in variable objects the emission lines follow the lines (e.g. Lyα,Hα,Hβ, CIII] λ1909), indicating a radially continuum tells us the clouds are at least optically thick stratified BLR with size ∼ 2–20 light days (figure 5). to the ionizing radiation, i.e. they are at least thicker S ∼ Uc/n α ∼ × 10 Surprisingly, these size estimates were a factor of ten than the Strömgren depth e B 5 10 n α smaller than previous estimates based on photoionization cm where e is the electron density and B is the case- − calculations (see below). The reason for the discrepancy is B hydrogen recombination coefficient (∼ 2.5 × 10 13 cm3 − now understood: not only does stratification undermine s 1). Matching the low ionization strengths requires the simple photoionization estimate but the densities in the extended partially ionized region to be a few times the BLR are probably higher than previously thought (e.g. deeper—perhaps a few times the diameter of our Sun. ∼ 1011 cm−3 in the C IV λ1459 emitting region). There is also evidence, it should be said, for an optically Since the ionization structure in the BLR is roughly thin component to the BLR gas, coming from studies of the same from object to object (because the line ratios are line ratio and profile variability. similar), we infer that BLR sizes must increase for more It is possible to estimate some other properties of the luminous objects. Using the C IV λ1549 data from NGC BLR line-emitting gas, starting with the basic equation R ∼ β L = 5548 as fiducial, the form of the dependence is BLR for the production of H by recombination: Hβ

Copyright © Nature Publishing Group 2001 Brunel Road, Houndmills, Basingstoke, Hampshire, RG21 6XS, UK Registered No. 785998 and Institute of Physics Publishing 2001 Dirac House, Temple Back, Bristol, BS1 6BE, UK 8 Seyfert Galaxies E NCYCLOPEDIA OF A STRONOMY AND A STROPHYSICS hν n n α V hν β Hβ e H Hβ where Hβ is the energy of each H Narrow line region n n V photon, e and H are the electon and proton densities, In the spectra of Seyferts, the narrow forbidden and α is the volume of ionized gas, and Hβ is the recombination permitted emission lines come from a region well outside coefficient for Hβ (∼ 3 × 10−14 cm3 s−1). Since the mass the BLR, spanning a few tens of pc to a few kpc called the M = m n V m of ionized gas is simply p H where p is the ‘Narrow Line Region’ (NLR). Unlike the BLR, therefore, proton mass, we find an ionized gas mass and volume of ground-based telescopes can often resolve the outer parts ∼ 0.1 M and 1045 cm3 for a typical Hβ luminosity ∼ 1042 of the NLR while HST has provided some detailed images erg s−1 and BLR gas density ∼ 1011 cm−3. Taking the BLR and spectra. The mere presence of forbidden lines tells radius of 10 light days from the variability measurements us that gas densities are lower than in the BLR. At these given above, we find that the ionized gas occupies only lower densities the physics of line production is simpler f ∼ . × −5 f c 1 5 10 of the BLR volume ( c is called the filling and line ratios may be used to estimate physical properties factor). If we assume the gas is in the form of clouds with of the ionized gas. Classic examples include the two λλ / λλ / diameters a few times the Strömgren depth (d ∼ 5 × 1011 line ratios: [S II] 6717 6731 and [O III] 5007 4363 c − N ∼ V/d3 ∼ 10 ∼ 3 4 3 cm) we derive a number of clouds c 10 which which yield electron densities 10 –10 cm and electron c ∼ × 4 collectively intercept a fraction of the central radiation of temperatures 1–2 10 K respectively. Such simple C ∼ N d2/ πr2 ∼ . C estimates must be viewed with caution, however, because c c c 4 0 3( c is called the covering factor). In this picture, then, the BLR is sparsely filled by a large they represent weighted averages over a wide range number of small clouds which collectively intercept ∼ 30% of conditions: for example, gas densities probably rise ∼ 2 −3 ∼ 6−8 −3 of the ionizing radiation. from 10 cm on kpc scales to 10 cm in the The velocity field in the BLR is poorly understood, inner regions. More sophisticated approaches follow the despite the fact that the emission line profiles clearly have same strategy as for the BLR: photoionization calculations Doppler origin. Profiles have approximately logarithmic evaluate conditions within a single gas cloud and, in form (flux ∝ log λ) often with slight asymmetry and/or more recent work, integrate over a distribution of cloud properties. As with the BLR gas, much of the NLR gas substructure, with full widths at half maximum (FWHM) − is thought to be optically thick to the ionizing radiation, anywhere from 1000 to 10 000 km s 1 and maximum − although studies of the He II λ4686/Hβ ratio indicate the wing widths of up to 30 000 km s 1. Lines in the same presence of some optically thin gas. In trying to match object can have slightly different shape (e.g. He II λ4686 these calculations to observations, sequences of models is often broader than Hβ), indicating both velocity and are evaluated with a range of radiation parameter and/or ionization stratification in the BLR, with higher velocities relative mix of optically thick to optically thin gas. Typical at smaller radii. One approach has been to fit each values for the radiation parameter found for the NLR are line by the sum of two components, one broad and one U ∼ 0.01, with some evidence for higher values in Seyfert very broad with possible blueshift. These fits support 1s than Seyfert 2s. the decrease in velocity for lower ionization gas, and Repeating the analysis for the BLR, we can derive point to a slight radial velocity component, although crude estimates for the global properties of the NLR gas. profile variability studies suggest that radial flow is not a L ∼ 41 −1 ∼ 3 Takingas fiducial, Hβ 10 erg s , a gas density of 10 dominant component of the velocity field. If the velocities −3 ∼ 2 cm and NLR size of 100 pc, we obtain a mass of ionized are gravitational, estimates of central mass M ∼ rv /G ∼ 7 −2 gas of ∼ 10 M with filling factor ∼ 10 . The Strömgren 7 8 M 10 –10 with some evidence for a correlation between depth in the clouds is ∼ 1018 cm; there are ∼ 106 clouds central mass and luminosity. The luminosities derived with integrated covering factor ∼ 1. While the numbers ∼ amount to 1% of the Eddington luminosity, which is the are far from precise, they do indicate that the amount of natural upper limit to accretion-powered energy release gas is dynamically unimportant, that it only sparsely fills when outward radiation pressure just balances inward the region, and that it intercepts a significant fraction of gravitational force, limiting further accretion. the ionizing radiation. The geometry of the BLR is also poorly understood, As the name suggests, the emission lines from the with spherical or flattened cloud distributions as obvious NLR have ‘narrow’ Doppler widths, with FWHM ∼ 200– possibilities, including a thin but flared disk, although 800 km s−1 (a few lie outside this range). The emission the absence of very narrow lines expected for face-on line shapes are almost always more peaky than Gaussian, disks weighs against this last possibility. The origin and and most have an asymmetry with the base and wing evolution of the cloud system are also poorly understood. stronger and more extended on the blue side than the red Possibilities that have been explored include a two-phase (figure 6). This asymmetry tells us that, at least for the medium in which the clouds are in pressure balance higher velocity, more nuclear gas, there is a radial flow and with a hot intercloud medium; clouds may be confined part of the region is obscured by dust. If the far side of the by magnetic forces; clouds may form in cooling post- region is obscured by intervening dust, then the flow is shocked gas; clouds may even comprise the extended outwards (the visible near side is blueshifted). If the clouds atmospheres of giant stars. Each of these has its strengths themselves contain dust, and the line emission comes from and weaknesses, and we must await further work to clarify the side of the cloud facing the central source, then the the true situation. flow is inwards (we cannot see the clouds on the near side

asymmetry. This indicates that, like the BLR, the NLR is stratified with the higher velocity, more nuclear gas having higher density and ionization degree. It is conceivable, though not yet shown convincingly, that a continuous sequence of conditions connects the inner NLR to the BLR. The spatial structure of the NLR emission has been studied using emission line imaging, both from the ground and using HST. In this technique, images are taken through two narrow pass filters, one centered on the color of the emission line (ON band) and one centered on the nearby continuum (OFF band). By subtracting the OFF band from the ON band one is left with a pure emission line image. Furthermore, by dividing two emission line images such as [O III] λ5007 and Hβ one can generate a map of the gas excitation which clearly highlights AGN-related emission ([O III]/Hβ ∼ 10) and distinguishes it from star formation emission ([O III]/Hβ<3). Such images indicate that the NLR is often elongated, with an axis parallel to the Figure 6. The [O III] λ5007 emission line profile from the nuclear region of NGC 7674 provides an extreme example of an radio axis. Indeed, for those Seyferts with well developed asymmetric blue wing. A velocity scale, line median, instrument radio sources the NLR emission is closely related to the resolution and asymmetry function are also shown. radio source, with brighter emission either coincident with jets and jet bends, or surrounding and sheathing radio lobes, sometimes with a curved ‘bow shock’ appearance. since we look into their backs). Recent observations have Spectroscopic studies show higher velocities for this radio- shown that the optical and infrared forbidden lines have related emission, indicating strong dynamical interaction similar asymmetries, showing that the dust is still opaque and an overall bipolar outflow, but also including split even at these long wavelengths. The preferred scenario is lines near jets, possibly indicating lateral expansion away thick nuclear dust, possibly in a dense disk, blocking the from the jet. A good example of close radio/emission line far side and indicating an outflow. The outflow might be association is Markarian 78 (figure 7). driven by a nuclear wind or be related to the jets which The close association between line and radio emission give rise to the radio source. For those Seyferts with in the NLR raises the important possibility that the gas is relatively strong double or triple structured radio sources ionized, at least in part, by shocks arising in the interaction. it is often possible to see separate red- and blueshifted Although early calculations of slow shocks could only emission components close to the radio lobes, indicating match LINER-like spectra, more recent calculations of fast bipolar driven outflows. In these objects, the integrated (∼ 500 km s−1) shocks can match Seyfert-like spectra. line profile can be particularly structured and broad, with A key feature of these calculations is the production of base and core widths up to ∼ 1000 km s−1 (examples UV and x-ray radiation, both line and continuum, in include NGC 1068, MKN 3, MKN 78). For most objects, the hot post-shock gas. This radiation then photoionizes however, there is good evidence that the majority of the gas both behind the shock and in front of it, yielding NLR gas, as measured by the [O III] λ5007 profile core a spectrum quite similar to that observed in Seyferts. width (FWHM), is moving in the gravitational field of the At present, ‘photoionizing shocks’ and standard power bulge of the host galaxy. The evidence includes the fact law photoionization (possibly including optically thin that the profile peaks and cores are close to the rest frame gas) give an equivalently good match at least to the of the galaxy, and that there are good correlations of the optical lines, though there is some hope that UV lines [O III] λ5007 emission line width with (a) rotation velocity will help discriminate between the models, as long as the of the galaxy, (b) bulge luminosity and (c) bulge stellar uncertainties introduced by dust reddening are not too velocity dispersion. Stated simply, galaxies with massive great. bulges have relatively broad forbidden lines (e.g. ∼ 500 km Returning to the emission line images and excitation s−1), while galaxies with low mass bulges have relatively maps, a number show sharp edges which outline parts narrow forbidden lines (e.g. ∼ 200 km s−1). Galaxies with of a cone, or bi-cone if it occurs on both sides. As relatively strong double or triple radio sources tend to described earlier, this seems to indicate a partly shadowed stand off these correlations with yet broader lines, due to nuclear source of ionizing radiation. These emission bipolar flows. For the gravitationally dominated gas, it is regions can extend out to considerable distances, ∼ 15– still unclear whether the predominant motion is rotation 20 kpc, where they are termed the ‘Extended Narrow or random, though probably both occur. Line Region’ or ENLR. This emission has high excitation Not all emission lines have the same profile shape, and narrow lines of low velocity, suggesting it is simply and in general lines of higher ionization and/or higher undisturbed ambient gas residing either in the disk of the critical density tend to be broader with a higher degree of galaxy or in the galactic environment. The fact that most

Hosts, environment and triggering While the previous sections have concentrated on the anatomy of activity in Seyfert galaxies, there remains an important question: why are some galaxies active while others are not? Although we cannot yet answer this question with confidence, two or three lines of enquiry help pave the way forward. The first is to ask whether Seyfert host galaxies are unusual in any way; the second is to ask whether any active properties correlate with any host galaxy properties; and the third is to identify mechanisms which might send gas down to fuel the nuclear regions. In this area, empirical studies are difficult because of the statistical weakness of the effects and the presence of sample and control biases. Perhaps the one clear result concerns the morphological type of the host galaxies: Seyferts favor early-type spirals, S0–Sbc, clearly avoiding late-type spirals and, to a lesser extent, ellipticals. Interestingly, LINERs show a similar distribution (with somewhat more ellipticals), while radio galaxies are almost exclusively elliptical or S0. In all these cases, more luminous galaxies are more likely to be active. It seems, therefore, that an important condition for the onset of activity is a relatively massive bulge. There is also a tendency for active galaxies to be disturbed in some way and/or have a nearby companion—both being conditions which may lead to nuclear fueling. The neutral gas content (atomic and molecular) of Seyferts is basically normal, though Seyfert 2s may have more molecular gas and associated star formation than otherwise similar non-active galaxies. Recent HST images have also shown Figure 7. HST [O III] emission line image and VLA radio image of the Seyfert 2 galaxy Mkn 78 showing close spatial association. that Seyfert 2s tend to inhabit later Hubble types and have Note that a dust lane crosses the nucleus, hiding the most more nuclear dust lanes than Seyfert 1s. This suggests a nuclear [O III] emission. Gray scale numbers are proportional to modification to the standard unification scheme: Seyfert flux, with the radio core rescaled by a factor of five for clarity. 2s simply have more nuclear dust than the Seyfert 1s and ∼ h = −1 −1 For this galaxy 1 arcsec 1 kpc ( 0 50 km s Mpc ). (Data are therefore more likely to be obscured, rather than being courtesy of M Whittle and A S Wilson.) identical except for our viewing angle. It is worth noting a few active properties which do (or don’t) correlate with host galaxy properties. First is ENLRs lie well beyond any radio source and can show the axis of nuclear activity, as defined by the radiation conical form suggests they are ionized by the central UV field and/or radio source. Overall, there appears to radiation source. Recently, however, careful estimates of be little alignment of this axis with the large scale disk the required UV flux suggest that at larger radii there may or bar of the host galaxy. Recent evidence, however, be an additional source of ionization, though what this is suggests that for later Hubble types there is an alignment is not yet known. with the host disk, while for early Hubble types the Afinal pragmatic theme is the frequent need to correct alignment is with smaller scale nuclear gas or dust disks or lanes whose more random orientation probably reflects measured spectra for the effects of reddening by dust. recent merger or infall history. Because the NLR resides Fortunately, the NLR Balmer line strengths are expected within the galaxy bulge, it is perhaps not surprising that to be close to their theoretical values, and so any deviation some NLR properties correlate with bulge luminosity, for can be attributed to reddening. For example, the Hα/Hβ ∼ . example emission line and radio luminosities. The origin flux ratio should be 2 8 (the ‘Case B’ value, though a of these correlations is not yet clear, but might involve . more detailed calculation gives 3 1) and so a measured a number of properties which depend on bulge mass R value of, say, , yields an estimate for the absorption (i.e. depth of gravitational potential): emission region A . (R/ . ) due to dust of V 6 8 log10 2 8 magnitudes in the V volume, gas and radio source pressures, mass loss rate band, where the coefficient depends on the reddening law from bulge stars, and mass of central black hole. More A (Whitford’s law in this case). Knowing V allows one to easy to understand is the dependence of NLR emission line correct the observed spectrum, using the same reddening width on bulge mass, since the gas velocities are largely f (λ) A = A × f (λ) law , where λ V . gravitational. An interesting wrinkle on this last result is

Copyright © Nature Publishing Group 2001 Brunel Road, Houndmills, Basingstoke, Hampshire, RG21 6XS, UK Registered No. 785998 and Institute of Physics Publishing 2001 Dirac House, Temple Back, Bristol, BS1 6BE, UK 11 Seyfert Galaxies E NCYCLOPEDIA OF A STRONOMY AND A STROPHYSICS that disturbed galaxies have even broader emission lines, Clearly, these are exciting times in the study of confirming that such disturbances can affect gas motions in Seyferts. the inner regions. Perhaps more fundamental is a possible correlation between nuclear (active) luminosity and host Bibliography galaxy (or more likely, bulge) luminosity—low luminosity Some excellent graduate level texts include: Seyferts reside in lower mass bulges than quasars. An Osterbrock D E 1989 Astrophysics of Gaseous Nebulae and obvious explanation is that more massive bulges have Active Galactic Nuclei (University Science Books) more massive central black holes, a result supported at Peterson B M 1997 An Introduction to Active Galactic Nuclei least in part by recent black hole mass measurements in (Cambridge: Cambridge University Press) more normal galaxies. Krolik J H 1999 Active Galactic Nuclei, From the Central Perhaps the central theoretical problem has been to Black Hole to the Galactic Environment (Princeton, NJ: understand how gas moves from far out in the galaxy disk Princeton University Press) down to the innermost regions where it can accrete onto the 5 Kembhavi A K and Narlikar J V 1999 Quasars and black hole—a loss in ANGULAR MOMENTUM by a factor ∼ 10 ! Active Galactic Nuclei, An Introduction (Cambridge: Early work focused on bars as a means of draining angular Cambridge University Press) momentum from gas, via shocks and dissipation. It is at present unclear how important this particular mechanism Some useful review articles in approximate chronological is since, although studies have been somewhat ambiguous, order include: there seems to be no tendency for active nuclei to prefer barred galaxies. More recently, Barnes and Hernquist Weedman D W 1977 Seyfert galaxies Ann. Rev. Astron. have shown, using computer simulations, that galaxy Astrophys. 15 69–95 encounters and mergers can be particularly effective at Balick B and Heckman T M 1982 Extranuclear Clues to the sending gas into the nuclear regions. The tidal fields Origin and Evolution ofActivity in Galaxies Ann. Rev. resulting from the encounter help create a gaseous bar Astron. Astrophys. 20 431–68 which lags behind a stellar bar. The gravitational torque on Rees M J 1984 Black hole models for active galactic nuclei the gas robs it of angular momentum and it falls inwards, Ann. Rev. Astron. Astrophys. 22 471–506 losing up to ∼ 99% of its angular momentum. At smaller Osterbrock D E and Mathews W G 1986 Emission line radii it is unclear what happens: gas may build up and regions of active galaxies and QSOs Ann. Rev. Astron. become unstable to self gravity and/or there may be bars Astrophys. 24 171–203 within bars which help funnel the gas to ever smaller radii. Osterbrock D E 1991 Active galactic nuclei Rep. Prog. Phys. Once in an accretion disk, however, Balbus and Hawley 54 579–633 have shown that magnetically driven turbulence provides Barnes J E and Hernquist L 1992 Dynamics of interacting enough viscosity to drive the gas inwards to the black hole. galaxies Ann. Rev. Astron. Astrophys. 30 705–42 Antonucci R 1993 Unified Models for active galactic nuclei The future Ann. Rev. Astron. Astrophys. 31 473–521 We now briefly identify several themes likely to be Mushotsky R F, Done C and Pounds KA1993 X-ray spectra pursued in the coming years. (a) The hunt for black and time variability of active galactic nuclei Ann. Rev. holes will continue, aiming to establish a demographic Astron. Astrophys. 31 717–61 perspective—how do black hole mass and spin relate to Peterson B M 1993 Reverberation mapping of active AGN or host properties? (b) How does the inner accretion galactic nuclei Publ. Astron. Soc. Pacific 105 247–68 disk function, and how does it play a role in generating the Koratkar A and Blaes O 1999 The ultraviolet and optical continuum emission? (c) Much remains to be done to find continuum emission in active galactic nuclei: the out the structure and velocity field of the BLR—ideally, its status of accretion disks Publ. Astron. Soc. Pacific 111 kinematics will be dominated by black hole gravity giving 1–30 another handle on the properties of the central object. (d) The root cause of the apparent unification of Seyfert 1s Mark Whittle and 2s needs to be clarified—is the torus model too simple in the light of the recent HST observations showing more distributed dust lanes in Seyfert 2s? (e) We need to refine our picture of the NLR, and in particular the importance of the radio emitting jets—clearly the jets accelerate the gas, but do they also ionize it? (f) What is the nature of the extended UV continuum in Seyfert 2s, and is enhanced star formation associated with nuclear activity? (g) Finally, we would like to know how activity is triggered in more detail. Although interactions are now known to send gas to the central kpc, how does it get from there to the pc scale and beyond?